Dear all,
I have been using the TFCE toolbox developed for SPM (I am using SPM12) and I have encountered some problems. I am using diffusion images (DTI and NODDI) as input and I have a flexible factorial design computed from CAT12, with two groups (12 healthy controls and 18 Multiple Sclerosis patients). When I try to use TFCE with this design, I set the number of permutations to 5000 but this number is never reached, and the number of possible permutations decreases to 4096. Furthermore, I get a warning saying "WARNING: Large discrepancy between parametric and non-parametric statistic found (cc=0.471878)! Probably your design was not correctly recognized", but the design is recognised as a repeated measures ANOVA, which seems fine. Finally, I computed an F-contrast to investigate interactions between Group and Time, and when I analysed the interaction clusters from TFCE, I noted that some of them fell outside the diffusion image, which seems strange. I also computed this contrast using SPM and everything seemed fine.
Does any of this happen because of my data? Or is there something else?
Thank you,
Maria Caranova
|